Information retrieval (IR) technology is widely used in search engines for web searches. Generally, inverted indexing is used to increase speed of a retrieval algorithm. However, as data scale becomes large, scanning candidate items in an index table can become quite time consuming. Such time demands can affect tasks such as retrieval of relevance ads. For example, a conventional system may need to retrieve relevance ads from tens of millions of ads in a time period of about 30 ms. Thus, a need exists for more efficient retrieval technology.